Agent Skills

Agent Skills are structured guidelines for AI coding agents. They provide reusable instructions that help an agent generate higher-quality code for a specific task, for example calling the Halfbill API correctly from a new project.

What Are Agent Skills?

Agent Skills are packaged sets of rules and guidelines that teach AI coding agents how to implement specific features correctly. A skill typically covers:

• API integration patterns
• Error handling strategies
• Performance and cost considerations

Claude Code Skills

Claude Code loads skills from a single SKILL.md file with YAML frontmatter. Claude Code activates a skill automatically when its description matches what you ask, or when you type its name.

A minimal skill that teaches the agent how to use Halfbill:

---
name: halfbill-api
description: Call Claude and GPT models through the Halfbill API.
---

When the project needs an LLM call, use the OpenAI SDK with
`baseURL: "https://api.halfbill.uk/v1"` and the `HALFBILL_API_KEY`
environment variable. Use `claude-sonnet-5` for Anthropic models and
`gpt-5.5` for OpenAI models. Never hardcode the API key.

Install a skill by placing its folder in your project's .claude/skills/ directory:

mkdir -p .claude/skills/halfbill-api
mv SKILL.md .claude/skills/halfbill-api/

Then trigger it by name inside Claude Code:

> halfbill-api

Place the folder in ~/.claude/skills/ instead to make the skill available across every project.

Using Skills with Other Agents

Most AI coding tools support custom instructions. Copy the skill content into your tool's configuration file:

With Cursor

Append the instructions to your .cursorrules file:

cat .claude/skills/halfbill-api/SKILL.md >> .cursorrules

With Other AI Agents

Tools that read an AGENTS.md file (Codex CLI, Continue, and others) pick up the same instructions when you add them to that file at the project root.